Bitcoin’s death spiral, a sea serpent as likely as the Loch Ness monster

At the end of this summer, a technical topic is being discussed in the French bitcoin community, the loss of profitability of mining after halving, supposedly leading to the “death spiral”. It is clear that the mechanisms of the Pow remain poorly understood, sometimes even within the community itself; Boursorama received in early July a speaker presented as a specialist in bitcoin, Philippe Béchade; he opened the interview with a jubilant “why did the fall of the BTC miraculously stop at $ 3100? Because that’s the Chinese production cost of a BTC, it couldn’t go any lower. The fact that Béchade doesn’t hear anything about PoW is anecdotal; there’s no question of contradicting him, he might not throw out such funny stuff anymore, that would be a shame. On the other hand, when Sisyphus, a respectable bitcoiner twittos questions the possible dangers of successive halvings and evokes in this regard “his own death in his code”, it shakes up the Slack of the Coin Circle, hence the idea of its members to make an article on the subject, which we hope will be definitive. I’m going to do it, so…

This theory, “the death spiral”, is a real sea serpent; we believe it has been definitively drowned in the shallows when it reappears, still as twisted as ever. This topic was debated as early as late 2011, came back in 2014-2015, and then was brought back into fashion in 2017 by Roger Ver, who went off on one of his delusions, imagining causing a death spiral on the Bitcoin network through the adoption of his Bitcoin Cash. The most outrageous version of the story was written by Atulya Sarin, a professor of finance at Santa Clara University, in an editorial entitled “Bitcoin is about to become worthless. I reacted abruptly at the time, saying only that this obscure finance professor with a lack of publications would have been better off lying down rather than feeling the wings to attack the foundations of a brilliant computer architecture he didn’t understand. If the sequel obviously proved him wrong, I will not spare any effort this time to explain how it all works, in real life.

The idea raised by Sisyphus is that by halving, miners will no longer be profitable, and that calls into question the very future of bitcoin.

This is the famous death spiral:

In this scenario, a large number of miners decide to stop mining simultaneously, as the activity is no longer profitable for them. Bitcoin’s protocol rules are structured so that a new block of transactions is mined approximately every ten minutes. Blocks will be found more quickly if more miners create computing power on the network and less frequently when miners start leaving the network.

The basic idea behind this “death spiral” is that all it takes is for a lot of miners to decide to disconnect after a sharp drop in price (for example) for the network to be paralyzed and no new transactions to take place.

Does it make sense? Yes, quite, if we ignore the fact that there is a mechanism for adjusting the difficulty every 2016 blocks, or about every 15 days at the normal rate. This adjustment will make it easier to find blocks if the Hash rate drops from the previous 2016 blocks. This means that this hypothetical death spiral only has this amount of time, the time to add 2016 blocks to the chain, to grow and kill bitcoin. So, of course, if the hashrate drops drastically over a few days, the time to add blocks gets longer, so the 2 weeks will quickly become 4, 6 or 8 weeks. We know that the sharpest downward correction for mining difficulty in the ASIC era was 15%, that was in December 2018. This shows that even in the midst of the bearmarket, the hashrate fall is slow. But let’s face it, a drop spiked with magic potion, a supernatural drop; let’s assume an extreme scenario, then, with 75% hashrate drop between two difficulty adjustments. Then a block would be extracted every 40 minutes rather than every 10 minutes. If the drop occurred at the beginning of the new difficulty period, then it would take six to seven weeks from the last difficulty adjustment for it to change again. This is not enough time to kill the grid: miners pay their electricity in advance, one month for the better off, three months for others. So, for 90 days for many miners, whatever the reward, the work will continue. For those who produce their own juice, and for those who steal it, it will continue indefinitely. So, there, end of the spiral…

Furthermore, in such a scenario, transaction costs would have skyrocketed throughout the decline due to the slower addition of blocks. This means that the decline in mining profitability could be offset by an increase in transaction costs, which would then rebound.

This mechanism is very well summarized by LUDOM, member of the Cercle du Coin on the Slack of the association: “So for me, halving is not a problem. The difficulty will adapt gradually. Even if the computing power is halved all at once, it only means that the frequency of the blocks will be 20 minutes while the difficulty adapts. This means that in the worst case there will be a small momentary bottleneck in transactions. And people will have to raise fees to get their transactions through the bottleneck (which will increase the profitability of mining those blocks).”

In the past, there have been times when the transaction fee has been a larger part of the overall reward than the grant (the newly created bitcoin). The last four times this happened was during the bubble’s deflation in late 2017. There, the conditions were right for a death spiral, according to proponents of this theory… it didn’t happen.

For the more pessimistic, a huge, worldwide and simultaneous increase in KWh prices could create the conditions for such a spiral; but even in this case there would be a solution: the problem could be solved by changing Bitcoin’s difficulty adjustment algorithm via a hard fork. It wouldn’t cause a split, everyone would move to the new network with their revised difficulty adjustment method, mining on the old chain would be pointless.

Well, here we are in science fiction, this fork will not happen; but it is a question of envisaging the worst, our community likes to scare itself, and it is healthy.

It should also be remembered that when some mines cease operations, those that remain become more profitable as soon as the next adjustment in difficulty occurs. In other words, a miner’s break-even point goes from $6,000 to $3,000 if half the hashrate disappears. Then from 3,000 to 1,500, without any problem (this is what our friend Béchade did not understand).

It should also be noted, as Julien Guitton (Weedcoder) reminds us, that “mining is an activity that requires a lot of infrastructure. miners try to make enough money during bull markets to survive bear markets. the less prepared disappear or are bought out. Also, even though it is fairly calcified, the Bitcoin protocol is still subject to evolution. Things like Segwit, Schnorr or other block size optimization allow for optimization of the number of transactions per block and thus an increase in fees.”

To which Jean Luc of bitcoin.fr adds: “Optimizations yes… but the number of transactions per second will not be multiplied by 100. At the next ATH, the Lightning Network may become unavoidable.”

No doubt that the Lightning Network would not be essential from this halving of 2020, but LN exists, works and brings us the certainty that in the medium term, each block will be able to receive many, many more transactions (100 k/second according to the specialist Béchade), guaranteeing in the long run an income to the miners via the fees.

So, when we approach 2140, when no more new tokens are generated, mining will not be ruined and bitcoin will not disappear. André Stilman, administrator of the Cercle du Coin, reminds us that Satoshi had mentioned the subject in his rare writings: “White paper SN. Point 6 (but you have to read between the lines…). ‘initial distribution of currency’ and then ‘may pass on… transaction fees’; It’s been working like this for 10 years and two halvings. I think the biggest risk was in the first halving. Now that the machine has been set in motion, it’s not really overly optimistic.”

Obviously, we won’t be around to see that bitcoin performs in 2140, but it is important to understand that we miners are building a network that we know is built for the long term. Few of us would put so much energy into this competition as to imagine that it could stop because of a lack of consistency in the model; we have definitely excluded that from the list of threats to our operations. Satoshi has done an extremely good job, his protocol stands the test of time: bitcoiners, don’t be afraid of the Loch Ness monster, it’s a fake.